Sophie
LEDUCQ

Dermatologic diseases might include one, several or multiple cutaneous lesions. They might be treated with systemic or topical drugs, according to the number of skin lesions. Different possibilities for randomized controlled trials (RCTs) assessing topical treatments can be considered. Randomization units may be the patient or his lesions; all lesions may be assessed or only one. Therefore, a classical individually RCT, a cluster RCT (in which clusters are patients, and units within clusters are lesions), or a within-person RCT might be designed. The objective of this work is to investigate methodological issues regarding RCTs assessing topical drugs in dermatologic conditions. First, we will identify and discuss key methodological issues for each design. Second, we will perform a systematic review of RCTs on topical drugs in dermatology that used a cluster design or a within-person design. Third, we will perform statistical calculations (Monte-Carlo simulations) for the study of statistical power in these 2 kinds of designs. Last, we will study the acceptance of patients and physicians for these 3 designs by using revealed preference methods.